ActaBIOMATERIALIA

Acta Biomaterialia 1 (2005) 183–191 www.actamat-journals.com

A stable three creatinine biosensor. 2. Analysis of the impact of silver ions on creatine amidinohydrolase

Jason A. Berberich b,1, Lee Wei Yang a, Ivet Bahar a, Alan J. Russell b,*

a Center for Computational Biology & Bioinformatics and Department of Molecular Genetics & Biochemistry, School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA b Department of Surgery, McGowan Institute for Regenerative Medicine, University of Pittsburgh, Pittsburgh, PA 15219, USA

Received 11 October 2004; received in revised form 26 November 2004; accepted 28 November 2004

Abstract

The enzyme creatine amidinohydrolase is a clinically important enzyme used in the determination of creatinine in blood and urine. Continuous use biosensors are becoming more important in the clinical setting; however, long-use creatinine biosensors have not been commercialized due to the complexity of the three-enzyme creatinine biosensor and the lack of stability of its components. This paper, the second in a series of three, describes the immobilization and stabilization of creatine amidinohydrolase. Creatine amidinohydrolase modified with poly(ethylene glycol) activated with isocyanate retains significant activity after modification. The enzyme was successfully immobilized into hydrophilic polyurethanes using a reactive prepolymer strategy. The immobilized enzyme retained significant activity over a 30 day period at 37 °C and was irreversibly immobilized into the polymer. Despite being stabilized in the polymer, the enzyme remained highly sensitive to silver ions which were released from the amperometric electrodes. Computational analysis of the structure of the using the Gaussian network model suggests that the silver ions bind tightly to a cysteine residue preventing normal enzyme dynamics and catalysis. Ó 2004 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.

Keywords: Biosensor; Creatinine; Polyurethane; Creatine amidinohydrolase; Immobilization

1. Introduction in polyurethanes [4–6]. We hypothesize that the signifi- cant stabilization that can be imparted via incorpora- Clinical analyzers require biosensors to be used tion into polyurethanes will enable the construction of repeatedly while in contact with biological fluids (blood enzyme-containing layers capable of multiple use in or urine) [1,2]. These systems often require a biosensor amperometric biosensors. to operate continuously at 37 °C for more than one The detection of creatinine levels in biological fluids week (greater than 500 samples) [3]. For many biosen- is an increasingly important clinical goal. This analyte sors, the stability of the enzyme layer limits the practical is used for the evaluation of renal function and muscle lifetime of the sensor. Our interest lies in the stabiliza- damage. There has been considerable interest in the de- tion of via multi-point covalent immobilization sign of biosensors specific for creatinine. A three-enzyme biosensor has been developed that converts creatinine to amperometrically measurable hydrogen peroxide * Corresponding author. Tel.: +1 412 235 5200; fax: +1 412 235 5290. (H2O2) [7]. Due to the complexity of the three-enzyme E-mail address: [email protected] (A.J. Russell). system, the development and commercialization of the 1 Present Address: Agentase, LLC, Pittsburgh, PA, USA. creatinine biosensor has been slow [8]. A number of

1742-7061/$ - see front matter Ó 2004 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. doi:10.1016/j.actbio.2004.11.007 184 J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191

O

O H2N C NH2

H2N C N CH2 C OH Urea + H2O O NH CH3 Creatine amidinohydrolase Creatine (Creatinase) EC 3.5.3.3 H3CCHNH 2 C OH

Scheme 1. Creatine amidinohydrolase catalyzed hydrolysis of creatine. methods are reported in the literature for the immobili- dialysis against 50 mM phosphate buffer at 4 °C [12]. zation of enzymes for use in creatinine biosensors and Enzyme activity following modification was determined the method of immobilization is the most important fac- using an end point assay (described below). tor in the determination of the operational and storage stability of the biosensor [8]. 2.3. Synthesis of creatine amidinohydrolase-containing Creatine amidinohydrolase (creatinase, EC 3.5.3.3) is polyurethane hydrogel a homodimer with subunit molecular weights of approx- imately 45 kDa. The enzyme catalyzes the hydrolysis of Hypol prepolymer 2060G (0.4 g) was added to a buf- creatine (see Scheme 1). The two active sites of the pro- fered solution (3.6 g of 50 mM phosphate buffer, pH 7.5) tein are at the interface of the monomers being shared containing creatine amidinohydrolase (0–100 units en- by each monomer and only the dimer is active. The en- zyme per gram of prepolymer). The aqueous polymer zyme has a low functional stability [9]. Additives such as solution was mixed for 30 s in a weigh boat until the reducing agents, and polyols have been shown onset of gelation. Polymerization was very rapid and to increase the stability of the enzyme [9]. The low gelation was usually complete within 1 min. intrinsic stability of creatine amidinohydrolase has prompted the use of protein engineering to improve stability [10]. Herein we describe the modification, 2.4. Characterization of enzyme modification immobilization and stabilization of creatine amidino- . MALDI-MS analyses were performed using a Perspective Biosystems Voyager Elite MALDI-TOF. The acceleration voltage was set to 20 kV in a linear 2. Materials and methods mode. One microlitres of PEGylated enzyme solution (0.1 mg/ml) was mixed with 1 ll of matrix solution 2.1. Materials (0.5 ml water, 0.5 ml of acetonitrile, 1 ll of trifluoracetic acid, and 10 mg of sinapinic acid). Spectra were re- Creatine amidinohydrolase (from Actinobacillus sp., corded after evaporation of the solvent mixture and CRH-211) and sarcosine (from Arthrobacter were calibrated externally using equine cytochrome C sp., SAO-341) were purchased from Toyobo Co., Ltd. (12,361.96 Da), rabbit muscle aldolase (39,212.28 Da) All enzymes were used without further purification. and bovine serum albumin (66,430.09 Da). Poly(ethylene glycol) activated with isocyanate (mPEG-NCO) (Mw 5000) was obtained from Shearwa- 2.5. End point assay for creatine amidinohydrolase ter Polymers Inc. (Huntsville, AL). Hypol 2060G pre- activity polymer, a toluene diisocyanate based prepolymer [11], was purchased from Hampshire Chemical (Lexington, Creatine amidinohydrolase activity was monitored MA). All other reagents were purchased from Sigma-Al- using a colorimetric assay that measures urea formation drich Chemicals (St. Louis, MO) and were of the highest from the hydrolysis of creatine [7]. An enzyme solution purity available. (0.1 ml) was incubated with a mixture (0.90 ml) of 100 mM creatine in sodium phosphate buffer (50 mM, 2.2. PEGylation of creatine amidinohydrolase pH 7.5) at 37 °C for 10 min. The reaction was stopped by adding 2.0 ml of a solution containing EhrlichÕs re- PEG-NCO was added at room temperature to a buf- agent (2.0 g of p-dimethyl aminobenzaldehyde in fered solution (50 mM phosphate, pH 7.5) containing 100 ml of dimethyl sulfoxide plus 15 ml of concentrated 3 mg/ml creatine amidinohydrolase. The ratio of PEG- HCl). The solution was incubated for 20 min at room NCO to enzyme was adjusted from 0:1 to 100:1. The temperature and the absorbance at 435 nm was mea- reaction was mixed for 30 min, followed by overnight sured. One unit was defined as the amount of enzyme J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191 185 that catalyzes the hydrolysis of 1 lmol of substrate per incubate for 5 min and residual activity was determined minute. using the end point assay.

2.6. monitor assay for creatine 2.10. UV–Vis absorption spectra of creatine amidinohydrolase activity amidinohydrolase

The initial rate of oxygen consumption was measured UV spectra were collected as previously described at 37 °C with a Clark oxygen electrode from Yellow [13]. In short, creatine amidinohydrolase (1 mg/ml) Springs Instruments (Yellow Springs, OH). The reaction was dissolved in 20 mM Tris (pH 7.5) with varying was initiated by adding an enzyme solution (1 ll) or an concentrations of silver nitrate (0–100 lM). In quartz enzyme-containing polymer (10–100 mg cut into small cuvettes, the UV absorption spectra of the enzyme solu- pieces) to 5.0 ml of substrate (100 mM creatine in tions were collected from 230 to 300 nm. Difference 50 mM phosphate buffer, pH 7.5 with at least 6 U/ml spectra were obtained by subtracting the spectra of the sarcosine oxidase). Before measurement, the assay solu- native enzyme (no silver) from the silver-ion-containing tion was allowed to equilibrate to 37 °C with air. Oxy- spectra. gen consumption was measured for 5–10 min. 2.11. Selection of a suitable template for GNM analysis 2.7. Thermostability creatine amidinohydrolase The crystal structure of Actinobacillus creatinase (Protein data base (PDB) code: 1KP0) was used to ana- Thermoinactivation of native and PEG-creatine ami- lyze collective dynamics via the Gaussian network mod- dinohydrolase was monitored at 37 °C in buffer (50 mM el (GNM) [14]. The CMS (carbamoyl sarcosine, a sodium phosphate, 2 mM EDTA, pH 7.5). The enzyme creatine analog) bound form of A. Bacillus creatinase concentration in all samples was 0.08–0.1 mg/ml. Sam- was used in order to determine the location of the resi- ples were removed periodically and assayed for activity dues in the . The CMS bound A. Bacillus using the end point assay. creatinase structure was computationally modeled by Enzyme polymer samples were cut into small pieces combining the CMS molecule in P. putida creatinase and added to buffer (50 mM phosphate, pH 7.5) and structure (PDB code: 1CHM) with the A. Bacillus crea- incubated at 37 °C. Samples were removed periodically tinase structure, followed by standard energy minimiza- and assayed for enzymatic activity using the oxygen tion using the Molecular Operating Environment electrode. package (MOE) [15]. The local energy minimization algorithms of MOE apply nonlinear function optimiza- 2.8. Inhibition of creatine amidinohydrolase by silver ions tion to the potential energy function to determine a nearby conformation for which the forces on the atoms Creatine amidinohydrolase (1.0 mg/ml) was incu- are zero [16]. bated in 20 mM Tris–HCl (pH 7.5) with silver nitrate (0–1 mM) at room temperature. Samples were removed periodically and assayed using the end point assay. In 3. Results and discussion order to test if creatine acted as a competitive inhibitor for silver inhibition, the end point assay was performed 3.1. Effect of PEGylation on enzyme activity and stability as above with silver nitrate added to the creatine solu- tion (0–100 lM AgNO in 20 mM Tris buffer). After 3 Prior to covalent immobilization, we were interested 10 min of incubation, the stop solution (p-dimethyl in determining the effect of chemical modification on benzaldehyde solution) was added. Control experiments creatine amidinohydrolase activity and stability. PEGy- showed silver nitrate had no effect on the color lation accurately mimics the first steps in polyurethane- development. based immobilization strategies and we therefore PEGylated the enzyme. Creatine amidinohydrolase can 2.9. The use of additives to prevent silver inhibition of be modified with reactive PEGs to a high degree without creatine amidinohydrolase significant loss of enzyme activity (data not shown). Modification of each monomer of the enzyme with an Creatine amidinohydrolase (1.0 mg/ml) was prepared average of 5 PEG chains led to a loss of only 30% in solution with either 50 mM EDTA, 50 mM EGTA, activity. 50 mM mercaptoethanol, 50 mM DTT or 50 mM cys- Although much of the enzyme activity was retained teine with 20 mM Tris, pH 7.5. Silver nitrate was added after modification, a significant decrease in enzyme to give a final concentration of 0–100 lM. After the stability was observed when the modified enzyme was addition of silver nitrate, the solutions were allowed to stored in buffer at 37 °C(Fig. 1). Native enzyme loses 186 J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191

100 300

250 80

200 60 150 40 100

20 Relative Activity (%) Relative Activity (%) 50

0 0 010203040 020406080 Time (days) Time (days)

Fig. 1. Stability of creatine amidinohydrolase as a function of number Fig. 2. Stability of creatine amidinohydrolase-containing polyure- of PEGs attached at 37 °C. Native enzyme (closed diamonds); one thane hydrogels stored in buffer at 37 °C. Rates were normalized to the PEG attached (closed triangles); three PEGs attached (closed circles); rate of oxygen consumption after the first day. Native enzyme (open five PEGs attached (closed squares). circles); Immobilized enzyme: 10 units/g polymer (closed triangles); 50 units/g polymer (closed squares); 100 units/g polymer (closed circles). less than 40% activity in buffer at 37 °C over 40 days; however, the modified enzyme showed a substantial de- crease in stability in solution. Since creatine amidinohy- and the rate of loss in observed enzyme activity was drolase is a homodimer with a low intrinsic stability [9] it dependent on the concentration of the enzyme in the is possible that chemical modification induces unfolding polymer. Polymers with high enzyme content showed in this delicate protein. Although PEGylation often pre- slower deactivation. Nonetheless, significant activity dicts the impact of polyurethane immobilization on was retained in all polymers for up to 80 days which is activity, the stabilization of proteins by immobilization more than long enough for use in a diagnostic biosensor. can be significantly different than the impact of The stability reduction caused by PEGylation was not PEGylation. observed in the immobilized enzyme. Given that only upon immobilization is the enzyme ‘‘locked’’ into a fixed 3.2. Immobilization of creatine amidinohydrolase in conformation these data support an intuitive expecta- polyurethane polymers tion of stability enhancements.

Biopolymers were prepared with enzyme concentra- 3.3. Effect of silver on enzyme activity tions of 0–100 units/g of polymer. The specific activities of the enzyme–polymers were directly proportional to Before being used in a functioning sensor, polyure- enzyme concentration (data not shown). This implies thane-immobilized enzyme must be applied to ampero- that the rates measured under these conditions are not metric electrodes that contain Ag/AgCl. Given the diffusion-controlled and reflect only the enzymatic reac- known propensity of silver ions to interact with proteins tion rate [17]. The average activity retention for creatine [3,18], the effect of silver ions on enzyme activity was amidinohydrolase in the polyurethane polymers was explored. Silver-induced inactivation becomes more of 28%. a concern if a sensor will be used in series with other sen- The immobilized polymers were tested for reusability sors for a long period of time since the enzyme will have by repeatedly assaying a single gel sample for eight cy- significant time to scavenge silver from solution. The cles with no loss of apparent activity (data not shown). enzyme was incubated with 5–100 lM silver nitrate As seen with other polyurethane chemistries, the enzyme and assayed for residual creatine amidinohydrolase was well immobilized within the gel and no leaching of activity (Fig. 3). Silver is a very effective inhibitor of cre- enzyme from the polymer to the surrounding solvent atine amidinohydrolase, completely inhibiting enzyme was observed [4–6]. activity even when the silver concentration is on the The storage stability of the enzyme in the polymer same order as the enzyme concentration. was determined by incubating enzyme–polymer samples In order to determine whether silver was acting as a in buffer at 37 °C and removing samples periodically to competitive inhibitor, differing concentrations of silver be assayed (Fig. 2). Interestingly, the activity of the and 90 mM creatine were added to the buffer. The en- polymerized enzyme increased during the first week of zyme (0.7 lM) was added directly to this solution storage. This increase in activity may be due to a change and after 10 min the concentration of urea in the system in polymer properties although the effect was not ob- was determined (data not shown). The presence of crea- served with related enzymes in these polymers. After tine did not slow enzyme inhibition. Silver must there- the first week, the enzyme activity began to decrease fore be binding at a location other than the active site. J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191 187

100 247 nm was apparent with increasing concentrations of silver. This is likely due to the formation of ligand- 80 to-metal charge transition (LMCT) bands. LMCT

60 bands caused by charge transitions that occur when ligands in proteins bind to metal centers, such as Ag 40 (I) [13]. Using Raman spectroscopy, Shen et al. showed that silver ions formed covalent bonds with sulfur atoms 20 Relative Activity (%) in human serum albumin (HSA) [13]. Since Ag (I) is a soft Lewis acid, it should have a high affinity for the soft 0 0 20 40 60 80 100 120 donor sulfur atom [19] and hence one can expect strong µ AgNO3 ( M) interactions between the cysteine residues of creatine amidinohydrolase and silver. Fig. 3. Silver-induced deactivation of creatine amidinohydrolase in Coll et al. [20] and Yoshimoto et al. [21] have shown solution. Incubation time: 5 min (closed circles); 15 min (closed squares). that the modification of sulfhydryl groups on creatine amidinohydrolase causes complete loss of activity. This is interesting since the Cys residues are distant from the It is important to note that the inactivation was not active site and have no known role in catalysis [20,22]. reversible. Dilution of the inhibited enzyme did not The loss of activity has been attributed to the possible regenerate activity and nor did dialysis overnight with prevention of domain motions by alkylation, thus lock- either EDTA or DTT (metal chelating agents). This sug- ing the enzyme into an inactive conformation [20]. gests that silver is bound tightly to the enzyme and is not likely to dissociate. 3.4. Collective dynamics of creatine amidinohydrolase Since the enzyme is effectively inhibited by silver ions at concentrations that could be released from electrodes To further investigate how Cys residues, and their in a clinical biosensor, it is of interest to develop meth- interaction with silver ions, might contribute to the func- ods to prevent or at least slow enzyme inhibition by sil- tion and stability of creatine amidinohydrolase, the ver. Silver can be effectively scavenged by a number of dynamics of the protein were examined with the GNM chelating agents. We investigated using 50 mM solutions [14]. The GNM is an elastic network model that has of EDTA, EGTA, DTT, mercaptoethanol, cysteine, been shown to predict the collective dynamics of pro- imidazole and polyethyleneimine (PEI) to sequester sil- teins. The model results are in close agreement with ver. Inhibition by silver ions was effectively prevented the temperature (B) factors from X-ray crystallographic by pre-incubation with thiol containing compounds experiments [14,23] and the free energy costs of H/D ex- (cysteine, mercaptoethanol, DTT) and prevented some- change [24]. This approach permits us to decompose what by PEI (Fig. 4). Hence, thiol-containing com- conformational motions into a series of orthogonal pounds may be effective in scavenging silver ions that modes, ranked by their associated frequencies. The leach from the electrode and may be useful in preventing modes with the slowest frequency (the slowest mode), enzyme deactivation. also referred to as ÔglobalÕ motions, usually reveal the UV/Vis spectroscopy was used to determine if any functional movements that engage the entire molecule major structural changes occurred upon binding of sil- [24–32]. The fastest modes, on the other hand, indicate ver to the enzyme. An increase in absorbance at ÔlocalÕ motions, and point to individual residues involved in the early folding/stabilization process [28,32]. The major utility of the GNM is its efficient applicability 120 to the dynamics of large structures (such as creatine amidinohydrolase, a dimer of N = 804 residues) that 100 are beyond the range of molecular dynamics simulations. 80 The mean-square fluctuations of creatine amidinohy- Enzyme only drolase residues predicted by the GNM were compared 60 EDTA EGTA Cysteine with the experimental B factors [33] (Fig. 5(a)). The re- 40 Mercaptoethanol DTT sults are displayed for only one monomer since both Imidazole monomers show almost identical fluctuation behavior. Relative Activity (%) 20 PEI The monomer structure is composed of two lobes, 0 N-lobe and C-lobe at the N- and C-termini, composed 0 20406080100 of 160 and 240 residues, respectively. The N-terminal AgNO (µM) 3 end shows the highest fluctuations. The close agreement Fig. 4. Protection of creatine amidinohydrolase from silver-induced between theoretical and experimental values further inactivation. validates the use of the GNM model. Fig. 5(b) displays 188 J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191

Fig. 5. Residue fluctuations in creatinase. (a) Comparison of predicted (by GNM; shown in blue) and experimental (X-ray crystallographic; red) B-factors. (b) Color-coded ribbon diagram of the monomer showing the most flexible regions (peaks in a) in red, and the least flexible (minima) in blue. (c) Ribbon diagram illustrating the relative positions of the monomers A and B with respect to the ligand, CMS, which is indicated by a black arrow. (For interpretation of color in this figure, the reader is referred to the web version of this article.) the ribbon diagram of the enzyme, color-coded from red to blue, 2 in the order of decreasing mean-square-fluctu- ations. Fig. 5(c) shows the location of the two mono- mers and the substrate analog (CMS). Next, we focused on the global motion modes to infer monomer A monomer B information on functional dynamics. Fig. 6 shows the mobilities of residues in the most representative global mode of the enzyme, which effectively displays the demarcation between the two lobes. The ordinate scales with the square displacements of the individual residues. Peaks indicate the most mobile regions, and minima re- fer to the global hinge (or anchoring) regions, about which the concerted motions of large substructures (do- mains, subunits, etc.) take place. The residues involved in the catalytic activity of the enzyme (shown by the Distribution of Mobilities in Global Motions open blue circles) typically exhibit minimal fluctuations. 0 100 200 300 400 500 600 700 800 This type of mechanical constraint near the catalytic site Residue Index is consistent with the fine-tuning and of Fig. 6. Distribution of mobilities in the dominant global mode of enzymatic activity, as observed in other enzymes [34]. creatinease dimer. The catalytic residues Phe62, Arg64, His231, We note that the substrate analog (CMS) is not symmet- Tyr257, Glu261, Arg 334 and Glu357 are shown by the blue open rically bound with respect to the two monomers in the circles and Cys60, Cys249 and Cys297 are shown by orange squares, in both monomers (separated by the dashed line). Red arrows indicate the catalytic residues that coordinate the CMS in the examined crystal 2 For interpretation of color in Figs. 5–8, the reader is referred to structure. (For interpretation of color in this figure, the reader is the web version of this article. referred to the web version of this article.) J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191 189

Fig. 7. (a) Color-coded ribbon diagram illustrating the mobility of creatinase residues in global motions. The colors blue–green–yellow–orange–red are used in the order of increasing mobility. Creatine analog, CMS, is shown in space-filling representation. The yellow arrows indicate the three cysteines (shown in ball-and-stick) on monomer B, and the white arrow the Cys60 on monomer A. Atoms in CMS and cysteines are colored by CPK convention. (b) Dynamics near the active site. Creatine analog, CMS, and key catalytic residues are shown in ball-and-stick. Catalytic residuesÕ sidechains and their associated backbone (ribbon) are colored according to their global mobilities (same as (a), shown from a different perspective for clarity). Hydrogen bonds are shown in dashed lines. The location of the peptide bond cleavage during catalytic electron transfer is shown by the white arrow. The moderate mobility of Glu261 and Glu357, which form hydrogen bonds with the nitrogen atom in the guanidine group of CMS may facilitate the peptide bond cleavage. Phe62 and Arg64 join the catalytic pocket from the other chain. (For interpretation of color in this figure, the reader is referred to the web version of this article.)

PDB structure, but interacts more closely with one of form a Ôhydrogen-bond likeÕ interaction with the nitro- the chains (monomer B) and subset of catalytic residues gen Nd1 on His 231, which could assist the His231 side (shown by arrows in Fig. 6). That said, both chains ex- chain in stabilizing a catalytically potent conformation. hibit the same global dynamics, and thus are equally dis- The folding nuclei were inferred from the high fre- posed to bind the substrate. quency modes predicted by the GNM (Fig. 8). Among The N-terminal lobe exhibits the largest amplitude the three cysteine residues, we note that Cys297 is distin- motions. Cys60 and Cys297 are both located in highly guished by a peak in Fig. 8. Cys297 can be predicted, constrained hinge/anchoring regions on the respective therefore, to be one of the most critical residues that C- and N-lobes (Fig. 6). In particular Cys297 resides are involved in the early folding process in creatine ami- in a highly stable central region near the interface be- dinohydrolase. We also observe a peak near Cys249. tween the two lobes, and plays a dual role in controlling domain and sub-domain motions (evidenced by the min- 0.09 ima observed at this residue in the dominant modes). I51 C297 I51 C297 L266 0.08 The N-terminal domain serves as a concertedly moving E190 E190 lid that allows the catalytic pocket in the C-terminal do- 0.07 main of the other chain to be exposed to solvent and to recruit the substrate inside. Cys60 belonging to the A 0.06 chain closely coordinates the substrate, and is also dis- 0.05 S173 tinguished by severely constrained and highly coopera- 0.04 tive dynamics (minima in Fig. 6). Fig. 7 provides a closer view of the catalytic binding 0.03

e1 e2 pocket. The oxygen atoms O of Glu357 and O of 0.02 Glu261 on monomer B form hydrogen bonds with a Participation in fastest modes nitrogen atom in the guanidine group of CMS (Fig. 7), 0.01 which may facilitate the peptide bond breakage between 0 the guanidine group and the resultant sarcosine mole- 0 100 200 300 400 500 600 700 800 Residue Index cule. Likewise, the residues Cys60, Phe62 and Arg64 on monomer A are spatially close to the B-monomer Fig. 8. Fluctuations in the high frequency modes. Peaks (labeled by His231. His231 is the most critical residue that dictates the residue types and numbers) reveal the centers of localization of the biocatalytic chemistry of creatine degradation energy subject to the highest frequency vibrations. Cys60, Cys249 and Cys297 are shown by red circles. The highest peak at Cys297 suggests [20,33]. His231 forms three hydrogen bonds with the that this residue significantly contributes to folding/stability. (For substrate (shown by dashed lines), but not with any interpretation of color in this figure, the reader is referred to the web c other amino acid. The sulfide atom (S ) on Cys60 could version of this article.) 190 J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191

Cys249 is located in the folding core composed of three References b-sheets (b9, b11 and b12) and part of two a-helices (a7 and a8) in a pita bread fold of C-terminal lobe. From [1] DÕOrazio P. Biosensors in clinical chemistry. Clin Chim Acta the electrostatic analysis, we also found that Cys297 is 2003;334:41–69. [2] Malhotra BD, Chaubey A. Biosensors for clinical diagnostics surrounded by negatively charged patches (data not industry. Sensors Actuators B 2003;91:117–27. shown). We hypothesize that these patches preferentially [3] Schaffar BPH. Thick film biosensors for metabolites in undiluted recruit silver ions to Cys297, which is both structurally whole blood and plasma samples. Anal Bioanal Chem 2002;372: and functionally critical. 254–60. Obviously, silver binds strongly with creatine ami- [4] LeJeune KE, Russell AJ. Covalent binding of a nerve agent hydrolyzing enzyme within polyurethane foams. Biotechnol Bio- dinohydrolase and completely inhibits enzyme function eng 1996;51:450–7. even at low concentrations. As reported in the other [5] LeJeune KE, Mesiano AJ, Bower SB, Grimsley JK, Wild JR, papers in this series, all three enzymes needed for the Russell AJ. Dramatically stabilized phosphotriesterase-poly- creatinine biosensor are inhibited to some extent by mers for nerve agent degradation. Biotechnol Bioeng 1997;54: silver; thus, methods for preventing or slowing enzyme 105–14. [6] Drevon GF, Danielmeier K, Federspiel W, Stolz DB, Wicks DA, inactivation by silver are important. Yu PC, et al. High-activity enzyme-polyurethane coatings. Bio- technol Bioeng 2002;79:785–94. [7] Tsuchida T, Yoda K. Multi-enzyme membrane electrodes for 4. Conclusions determination of creatinine and creatine in serum. Clin Chem 1983;29:51–5. We have shown that creatine amidinohydrolase can [8] Killard AJ, Smyth MR. Creatinine biosensors: principles and be effectively modified by PEG-NCO and covalently designs. Trends Biotechnol 2000;18:433–7. [9] Schumann J, Mo¨llering H, Jaenicke R. Intrinsic stability and incorporated into polyurethane materials via modifica- extrinsic stabilization of creatinase from Pseudomonas putida. Biol tion by isocyanate prepolymers. Modification by Chem 1993;374:427–34. increasing amounts of PEG appears to destabilize the [10] Schumann J, Bo¨hm G, Schumacher G, Rudolph R, Jaenicke R. enzyme leading to a decreased half-life when stored in Stabilization of creatinase from Pseudomonas putida by random buffer at room temperature. Creatine amidinohydrolase mutagenesis. Protein Sci 1993;2:1612–20. [11] Braatz JA. Biocompatible polyurethane-based hydrogel. J Biomat was successfully and irreversibly immobilized by chemi- Appl 1994;9:71–96. cal crosslinking into polyurethane polymers and re- [12] Drevon GF, Hartleib J, Scharff E, Ru¨terjans H, Russell AJ. tained significant activity (28%). Enzyme in the Thermoinactivation of diisopropoylfluorophosphatase-con- polymers was more than sufficiently stable to be used taining polyurethane polymers. Biomacromolecules 2001;2: for 30 days in a buffered solution at 37 °C. Creatine 764–71. [13] Shen X, Liang H, Guo J, Song C, He X, Yuan Y. Studies on the amidinohydrolase is highly susceptible to inhibition by interaction between Ag+ and human serum albumin. J Inorg silver. Inhibition does not appear to be caused by a per- Biochem 2003;95:124–30. turbation at the active site. Addition of silver scavenging [14] Bahar I, Atilgan AR, Erman B. Direct evaluation of thermal thiol-containing molecules was effective in preventing fluctuations in protein using a single parameter harmonic the loss of enzyme activity due to silver. Spectroscopic potential. Folding Design 1997;2:173–81. [15] MOE (The Molecular Operating Environment) Version 2004.03, studies suggest that metal-to-thiolate interactions may software available from Chemical Computing Group Inc., 1010 exist. Through the GNM analysis of the molecular Sherbrooke Street West, Suite 910, Montreal, Canada H3A 2R7. dynamics of creatine amidinohydrolase, we were able Available from: http://www.chemcomp.com. to predict that two of the cysteine residues, Cys60 [16] Gill P, Murray W. Practical optimization. London: Academic and Cys297 in both chains were critically important Press; 1981. [17] Yamane T. A proposal for accurate determination of ‘‘true’’ in controlling the functional dynamics of the enzyme. fractional retention of enzyme activity on immobilization. Bio- Furthermore, the analysis of high frequency motions technol Bioeng 1977;19:749–56. suggest, that, if modified, these two sites may destabi- [18] Karasawa Y, Takata Y. Maltose Sensor. US Patent 4,547,280, lize the enzyme and lead to inactivation. The fact that 1985. Cys297 is surrounded by negatively charged residues [19] Cowan JA. Inorganic biochemistry: an introduction. New York: VCH; 1993. gives further support to the potential attraction of pos- [20] Coll M, Knof SH, Ohga Y, Messerschmidt A, Huber R, itively charged silver near this residue. The use of crea- Moellering H, et al. Enzymatic mechanism of creatine amidino- tine amidinohydrolase in a biosensor will require hydrolase as deduced from crystal structures. J Mol Biol protection from silver ions. 1990;214:597–610. [21] Yoshimoto T, Oka I, Tsuru D. Purification, crystallization, and some properties of creatine amidinohydrolase from Pseudomonas Acknowledgment putida. J Biochem 1976;79:1381–3. [22] Hoeffken HW, Siegward HK, Bartlett PA, Huber R, Moellering H, Schumacher G. Crystal structure determination, refinement We gratefully acknowledge support provided by and molecular model of creatine amidinohydrolase from Pseudo- Bayer Diagnostics, Inc. monas putida. J Mol Biol 1988;204:417–33. J.A. Berberich et al. / Acta Biomaterialia 1 (2005) 183–191 191

[23] Kundu S, Melton JS, Sorenson DC, Phillips Jr GN. Dynamics of in relation to function and stability. Phys Rev Lett proteins in crystals: comparison of experiment with simple 1998;80:2733–6. models. Biophys J 2002;83:723–32. [29] Bahar I, Jernigan RL. Vibrational dynamics of transfer RNAs: [24] Bahar I, Wallqvist A, Covell DG, Jernigan RL. Correlation comparison of the free and synthetase-bound forms. J Mol Biol between native state hydrogen exchange and cooperative residue 1998;281:871–84. fluctuations from a simple model. Biochemistry 1998;37:1067–75. [30] Bahar I, Erman B, Jernigan RL, Atilgan AR, Covell DG. [25] Kitao A, Go N. Investigating protein dynamics in collective Collective motions in HIV-1 reverse transcriptase: examination of coordinate space. Curr Opin Struct Biol 1999;9(2):164–9. flexibility and enzyme function. J Mol Biol 1999;285:1023–37. [26] Ming D, Kong Y, Lambert MA, Huang Z, Ma J. How to describe [31] Tama F, Sanejou YH. Conformational change of proteins arising protein motion without amino acid sequence and atomic coordi- from normal mode calculations. Protein Eng 2001;14:1–6. nates. Proc Natl Acad Sci USA 2002;99:8620–5. [32] Rader AJ, Bahar I. Folding core predictions from network models [27] Haliloglu T, Bahar I. Structure-based analysis of protein dynam- of proteins. Polymer 2004;45(2):659–68. ics. Comparison of theoretical results for hen lysozyme with X-ray [33] Padmanabhan B, Paehler A, Horikoshi M. Structure of creatine diffraction and NMRrelaxation data. Proteins: Struct Function amidinohydrolase from actinobacillus. Acta Crystallogr Sect D Genetics 1999;37:654–67. 2002;58(8):1322–8. [28] Bahar I, Atilgan AR, Demirel MC, Burak E. Vibrational [34] Bartlett G, Thornton JM. Analysis of catalytic residues in enzyme dynamics of folded proteins: significance of slow and fast motions active sites. J Mol Biol 2002;324:105–21.